“The Lightweight Render Pipeline (LWRP) is a prebuilt Scriptable Render Pipeline (SRP) that is optimized for delivering high graphics performance. It’s scalable to mobile platforms, as well as higher-end consoles and PCs.”
At the moment I believe the built in Forward rendering path using the legacy diffuse lit shaders are still faster than the LWRP on most mobile platforms, but it’s pretty close and the LWRP is being actively worked on to make that better. Theoretically the performance should be at least identical or potentially faster for the LWRP. The legacy vertex lit path is faster yet, but that’s been deprecated.
Performance from best to worst currently looks roughly like this:
(Any) Unlit
(Built in Forward) Legacy/Vertex Lit
(LWRP) Baked Lit
(Built in Forward) Legacy/Diffuse
(LWRP) Simple Lit
(LWRP) Lit with specular & reflections disabled
(Built in Forward) Standard with specular & reflections disabled
(Built in Forward) Legacy/Specular
(LWRP) Lit
(Built in Forward) Standard
The differences between each depend on the content in the scene (number of lights, other factors like resolution and amount / size of textures, etc.) and the hardware being used. In some scenarios there may be no difference in measurable performance between even the Standard shader and the Unlit shader. In others the Simple Lit shader may outperform the Legacy/Diffuse, or the full Lit shader could outperform the Standard w/o spec & refl (that’ll be rare, but plausible).
And, like I said, right now you may see the LWRP performance being worse than the built in, but that should improve over time.
Static batching is still useful, and faster than SRP batching.
SRP Batching can be thought of as a faster dynamic batcher that takes some if the benefits of dynamic and static batching and removes the one thing in common between those two systems: SRP Batching does not combine meshes, only changes how data is organized on the GPU to reduce CPU usage.